Choose a model source
Start from the built-in provider directory or add a custom service compatible with OpenAI, Anthropic, or Ollama.
Jingle model connection
Bring your AI model into Jingle Desktop.
Jingle supports built-in model providers, custom OpenAI / Anthropic / Ollama-compatible endpoints, and local or keyless connections. You choose the model source; Jingle calls it only when you use AI features and stores credentials encrypted on your device.
Settings / AI models
Model connection center
Built-in providers
OpenAI, Anthropic, Google, and more
Custom endpoint
OpenAI / Anthropic / Ollama compatible
Default model
Switch per workflow anytime
Default model
Used by Assistant and Agents
Credential
Encrypted on device
Connection flow
First use follows one clear model setup path.
Connect models in the desktop app: choose a built-in provider, add a custom endpoint, or enable a local model; enter the required credentials; verify available models; then set the default.
Enter connection details
Add the API key, API URL, model names, or local address required by that provider. Local keyless services can be configured directly.
Verify available models
When you save, Jingle reads or confirms available models and shows a clear error if the connection fails.
Choose a default model
Set the model used by Assistant and Agents. You can switch by task whenever your needs change.
Trust boundary
You control the model connection; credentials serve only the AI features you authorize.
Choose your model source
Jingle does not lock you to one provider. Built-in providers, custom endpoints, and local models share one model settings surface.
Encrypt credentials locally
API keys and other sensitive fields are encrypted on your device. Model configuration and credentials are managed separately, and the full key is not exposed in the UI.
Inspect connection state
You can see which providers are ready, which model is default, and whether Jingle can read models from the connection.
Change or remove anytime
Update a key, edit an endpoint, change the default model, or remove a connection when your model strategy changes.
After connection
Once a model is connected, Jingle becomes your desktop AI workspace.
The default model powers Assistant, Agent tasks, and desktop flows that need model reasoning. You stay in control of provider choice, cost, speed, and context sharing.
AI Assistant
Ask questions, summarize context, draft content, and get answers from the default model you selected.
Agent workflows
Let Agents use approved tools, handle tasks, and show their work for you to inspect.
Local context
Work with visible files, memories, and task context while important access stays inspectable and controlled.
Model choice
Switch models for speed, cost, quality, reasoning depth, or team requirements without changing the rest of your workflow.
Safety principles
Clear, controllable, reversible.
You decide whether to use a built-in provider, custom endpoint, local model, and whether to provide an API key.
Jingle calls the model connection only when you run AI features; it will not use a model key to sign in to other accounts.
API keys and other sensitive values are encrypted on device and are not shown in ordinary UI or AI output.
Connection failures are shown clearly, and you can update, switch, or remove model connections anytime.